high throughput methodologies for chemicals and materials rd&e · the u.s. chemical process...

Invited Paper

High throughput methodologies for chemicals and materials RD&E1

by

John D. Hewes, Ph.D.National Institute of Standards and Technology

Advanced Technology ProgramChemistry and Life Sciences Office

100 Bureau Dr., Stop 4730Gaithersburg, MD 20899-473 0

I. SUMMARY

The U.S. chemical process industry (CPI) has annual revenues over $400 billion and supports the downstreaminnovations of other critical U.S. industries. Innovations in the chemical process industry support over $1 .2 trillion inproduct sales and service for the downstream value chains. Yet R&D funding within the chemical industry has beenrelatively flat over the last decade at about $12-14 billion while being focused increasingly on applied research andprocess development.2

Global market pressures are driving the chemicals and advanced materials industries to improve RD&E productivity asmeasured by return on investment for new products. The principal drivers are reduced cycle times and higher quality atlower cost, with reduced environmental impact. One approach to increased productivity is high throughput research (HTRD&E) (also known as "combinatorial chemistry"), where a convergence of technologies enables a parallel approach todiscovery and process development. These methodologies produce "libraries" of new solid-state and soluble materials ina matter of hours and days, rather than months and years, at a fraction of the cost of traditional serial approaches. Newmaterials have already been reported with unique chemical, optical, and electronic properties utilizing a variety ofparallel screening approaches.3

The CPI and materials industries have indicated a need to accelerate the implementation of this capability in the UnitedStates. Industry has indicated that the Advanced Technology Program (ATP) of the National Institute of Standards andTechnology (NIST) can facilitate the emergence of new technologies. The ATP can spur the fusion of these newtechnologies for subsequent diffusion to other areas. It can facilitate the integration of complex systems; the ATP canhelp the integration of otherwise diverse technologies; and the ATP can reduce the cost of implementing high throughputRD&E so that more R&D cost-sensitive industrial sectors will be able to invest in this new capability earlier. Thisposition paper is a response to industry's interest in the development of a portfolio of projects in the ATP to fundmethodology development in the area of high throughput RD&E for advanced materials (catalysts and biocatalysts,polymers and polymer blends, electronic materials, biomaterials, etc.) that could accelerate the building of synergies,promote inter-industry developments, and counter significant foreign competition in the CPI and materials sectors.

The ATP has responded to the need4 of industry for Federal Government investment in this area by awarding threeproject proposals from the FY 1999 Open Competition that competed successfully against the ATP Selection Criteria.These projects are attempting to develop breakthrough technologies for applications in polymer coatings andheterogeneous catalysts. The need for further growth has been expressed, and the ATP continues to encourageapplications for funding in future Open Competitions in the area of high throughput methods for RD&E of chemicalsand materials, as well as other high technical risk areas that have the potential for broad economic benefit for the nation.

This report is a collection point for non-proprietary business and technical information gleaned from industry input,for example as obtained at an Industry Probe Working Group Discussion held in March, 1998, public ATPWorkshops held November 18, 1998 in Atlanta, GA, November 16, 1999 in San Jose, CA, as well as ATP'sparticipation in many industrial forums in the last two years. Complete review articles can be found in thereferences.

Keywords: combinatorial chemistry, combinatorial methods, high throughput screening

In Combinatorial and Composition Spread Techniques in Materials and Device Development.46 Ghassan E. Jabbour, Editor, Proceedings of SPIE Vol. 3941 (2000) • 0277-786X1001$l 5.00

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 12/18/2015 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

NIST Special Publication 947This publication is available free of charge from:http://dx.doi.org/10.6028/NIST.SP.947

II. THE ATP PROCESS

The goal of the ATP is economic growth and the good jobs and quality of life that come with economic growth. ATPawards are made strictly on the basis of rigorous peer-reviewed competitions designed to select the proposals that arebest qualified in terms of the technological ideas, the potential economic benefits to the nation (not just the applicant),and the strength of the plan for eventual commercialization of the results. The ATP protects the confidentiality ofdocuments submitted by industry as required by the ATP statute. The ATP does not fund projects that are predominantlybasic research or product development. The ATP submits all proposals to peer review by following the process outlinedin Figure 1.

FULL PROPOSALS I*I SCREENING I

+

TECHNOLOGICAL AND ECONOMIC REVIEWTECHNOLOGICAL MERIT ECONOMIC MERIT

Innovations in the technology • Economic benefits• High technical risk & feasibility • Need for ATP funding• Quality of R&D Plan • Pathway to economic benefit

I[ HIGH TECHNOLOGICAL& ECONOMIC MERIT

SEMIFINALISTS

V

IDENTIFIED

FIGuRE 1 . THE ATI SELECTION PROCESS

Non-proprietary white papers (not proposals) submitted by industry, academia, and government facilities might outline aspecific opportunity area and describe the potential for U.S. economic benefit, the technical ideas available to beexploited, the strength of industry commitment to the work, and the reasons why ATP ftinding is necessary to achievewell-defmed research and business goals. These position papers contain no proprietary information and state assuccinctly as possible the goals of a proposed technology challenge that needs the benefit of a cluster of synergisticprojects to maximize benefit to the nation's economy.

ATP project proposals and optional pre-proposals submitted to AlP are evaluated against the scientific and technicalmerit of the proposal and the potential for broad-based economic benefits to the United States. A fifty-percent weightingapplies to the scientific and technological merits of the proposals: innovations in the technology; high technical risk andfeasibility; and quality ofthe research plan. A fifty-percent weighting is applied to the economic merits ofthe proposals:economic benefits; the pathway to economic benefit; and need for ATP funding. Overall, proposals are rankedcompetitively according to the six components of the two Selection Criteria as expressed in the written proposal.

Optional pre-proposals may be submitted year-round and provide a mechanism to receive written feedback on thestrengths and weaknesses of the project plan and on how to improve the competitive potential of a subsequent full

47

I DEBRIEFING IFINAL SELECTION Cooperative

Agreement


48

proposal. Applicants are strongly urged to initiate proposal preparations early and to follow instructions carefully.Proposal Preparation Kits for each new competition are distributed via the ATP Mailing List (telephone l-800-ATP-FUND to register) and on the AlP internet world wide web site http://www.atp.nist.gov/atp/apply.htm.

III. PROBLEM STATEMENT

Major forces—globalization of markets and the pace of technological change—continue to drive private sector R&D tomore focused, shorter-term investments to maximize near-term returns to the shareholders. Increased pressure onmanufacturers to produce "faster/lower-cost/better" has increased the demand for new products made from new materialsand/or utilizing new processes while utilizing more environmentally-friendly materials and processes. In order to achievethese goals, the CPI and materials sectors are expected to follow a trend occurring in other technology sectors and willfocus on core business areas with greater dependence on strategic partnerships. The move to outsource research anddevelopment efforts to outside firms (the "discovery and development industry") and/or create facilities in countries withlower-cost labor is an emerging business model in the CPI and materials sectors.

During the 1990's, the pharmaceutical industry responded to market pressures (product launch times, return on R&Dinvestment, large revenues for blockbuster drugs, reduced profits from health management organizations, etc.) byswitching from serial discovery processes for generating and analyzing drug targets to using "combinatorial chemistry"research methods. This methodology can qualify a large number (a "library") of possible candidates ("hits" from primaryscreening, "leads" from secondary screening); in 1999, the ability to screen 1 million distinct compounds per year wasrealized in the pharmaceutical discovery process, with R&D costs per lead dropping approximately two orders ofmagnitude. The capability to convert traditional discovery processes to an "assembly line" was facilitated by theconvergence of several technologies applied to a well-defmed market opportunity: computer software (data bases,molecular modeling, machine control software, and statistics); more cost-effective high-speed computers; robotics;MEMS (micro-electromechanical systems) technologies; and sensors.

Figure 2 illustrates how one can differentiate the key steps of a typical high throughput RD&E process:Target Defmition utilizing expert opinion, hypothesis, and computer modeling;Library Design with computational inputs such as quantitative structure-property relationships (QSPR), and molecularmodeling. Design ofExperiments (DOE) is required to reduce the number ofsamples that will be necessary to defmesample spaces within the experimental universe or to direct screening to other spaces within the universe;Library Fabrication involves the automated deposition and/or processing of an n-dimensional matrix of physical samples;Library Characterization in a parallel or massively parallel mode involves the use of robotics and sensors to rapidly andautomatically analyze the library of targets for desired properties;Data Collection and Analysis using the data-base and artificial intelligence tools—"informatics"—--expanded into themore complex realm of materials properties.

Figure 2. The High throughput Process Flow

Scale-up &Validate


25,000

FIGuRE 3. COMBINATORIAL VS. CONVENTIONAL REACTOR THROUGHPUT5

Industrial sectors other than the pharmaceutical industry recognized the attributes of the new drug discovery paradigmand are implementing high throughput screening of new materials. Serial discovery processes rely on the preparation andcharacterization of individual samples from bench scale (mg. or grams) to pilot scale (grams to kilograms). Parallel, HImethods can perform the same preparation and characterization using automated laboratory instrumentation. As shownin Figure 3, there are expectations that HI screening can significantly accelerate the discovery of new catalysts. Anemerging trend is to continue the parallel process flow into the process and product development stages to realizesubstantial increases in the number of samples that can be analyzed.

Many ofthe market factors that influenced drug discovery are presently driving a need for reducing the cycle time for thediscovery and process development of new advanced materials and lower-cost chemical products such as pharmaceuticalintermediates, fme and specialty chemicals, and commodity chemicals and materials. The technology spill-over fromdrug discovery has resulted in the development of high throughput (HI) methodologies for inorganic materials and fmeand specialty chemicals and materials. Because HI techniques are especially suited to complex mixtures containingmany different components and/or processing conditions, this methodology lends itself to the discovery of new advancedmaterials that are gaining performance through the use of multi-component formulations, for example the addition ofdopants to electronic materials or polymer blends for engineering plastics. In addition, HI RD&E permits the screeningof compositions that would not otherwise be attempted, that is, it maximizes serendipity. For example, SymyxTechnologies Inc.6 has reported the identification of a ternary fuel cell catalyst composition (M-M'-M") of highselectivity and conversion containing transition metals that show little or no catalytic activity in the three possible binarymixtures (M-M', M-M', M'-M"). Applications beyond catalysts have been explored, for example phosphors,thermoelectric materials, and optoelectronic devices such as ZnO-based compound semiconductors for light-emittingdiodes (LEDs).7

In the pharmaceutical industry, new drug candidate molecules are tested for biological activity by chemical means suchas gas chromatography, nuclear magnetic resonance spectrometry, mass spectroscopy, fluorescence spectroscopy, or theresponse of a biochemical receptor. Advanced materials, on the other hand, require performance-based characterizationthat is often application-specific. For example, selectivity and conversion analysis for a matrix of heterogeneous catalystformulations will require the development of sensors for sampling the product/reactant mixtures produced by the catalystlead sample array, not just an analysis for elemental composition or molecular activity.

Materials processing typically involves more energetic reaction environments than pharmaceutical processing, withtemperature and pressure requirements as high as several hundred degrees and hundreds of Pascals. Micro-scale solidstate samples may also be subject to significant influence from the substrate onto which they are deposited--interfacialeffects can produce phenomena that are not reproducible in bulk samples of manufacturing scale. Thus, advancedmaterials suffer from "scalability", or differences in micro-scale to lab- or pilot-scale properties. Finally, solid state

49

20,000

15,000

j 10,000

5,000

Time (Months)


50

compositions may develop into different (kinetically-controlled) metastable structures depending on processing andtesting conditions. Therefore, sample libraries must undergo validation at every process step.

The leverage of tools developed for drug discovery, e.g., activity-focused/solution-state systems, to solid state materialsis challenging due to the diversity of potential design, fabrication, and analysis parameters. Some of these challenges areshowninTable 1.

TABLE 1 . DRUG DISCOVERY TARGETS VS. SOLID STATE TARGETS

Drug Discovery vs. Solid State Materials

. Discrete molecules ofC, H, F, N, 0, P • Extended structures ofmany elementsnntentiallv in metastahie states

• Finite number ofactive sites, can be characterized Ill-defmed distribution ofactive sites andand modeled computationally stnictiires• Purity >85%; parallel purification techniques "Pure" solids are meaningless especially withemployed before characterization small samples having interfacial effects with the

library substrate

• Structures reproducible (a priori) Reproducible structures difficult, if notimpossible, to create a priori

• Chemical characterization, biological activity well • Characterization ofmultiple independentdeveloped for rapid or parallel methods properties and composition not straightforward• Descriptors for diversity Few or no ideas exist about how to do it

• Registration of library samples straightforward Few or no ideas exist about how to do it

• Synthetic building blocks available • A few building blocks available

Iv. TECHNOLOGY CHALLENGES

Information gained from the many public discussions8 in this area has been supplemented by additional input in the formof non-proprietary position papers submitted to the ATP by industry. In addition to identifying key areas of technologydevelopment and application areas, these data describe the technical and commercial barriers to implementing thismethodology in the global CPI and materials sectors.

The application of high throughput discovery and process development for chemicals and materials will drive theenabling integration of a hardware- and software-based infrastructure toward specific product applications. The long-term vision is to have high-throughput research become part of expanded enterprise-wide systems that include tools forhardware interfaces, technology assessment/decision, and logistics. Because HT RD&E is currently highly capitalintensive, with start-up costs in the range of $8M to $20M,9 discontinuous innovation in generic andlor modularhardware and software technologies will be necessary to drive down costs and facilitate its implementation in theindustrial sectors that have lower returns on R&D investment, such as exist in CPI and materials sectors.

A. SOFTWAREThe basic underlying software technology must be capable of defming profitable experimental spaces and visualizationof complex data relationships, and of correlating target materials with properties to permit data-base queries from a broadspectrum of data mining engines and the development of structure-property relationships. This requires interfacing withdata visualization tools at the back end and statistical experimental design engines on the front end while remainingcompliant with enterprise-wide systems for knowledge management and maintaining control ofexperimental hardware.The overall view of software integration is shown in Figure 4, where the integration oflnformatics, Modeling, andDesign are high-risk opportunities.

. Informatics. Integrated packages linking modeling, development and management of databases and search engines,hardware control, data visualization, and logistics. Data base search engines will need to be interoperable with thediverse flavors ofdatabases currently in use;


. Development of Quantitative Structure Property Relationships (QSPR) for materials. The term QuantitativeStructure Activity Relationships (QSAR) is used in organic combinatorial synthesis;. Acceleration ofthe design process from atomic level chemistry to engineering design by developingrelationships between chemistry, processing, microstructure, etc. and processing involving metastable states,etc.;

. Development of a query language for linking many different methods for querying the data with appropriatequery optimization methods;. Assembly of a high-performance data mining toolbox that extends a database management system withadditional operators;

S Connection to the diverse metrics in materials design, where important properties are sensitive to numerousranges of length or time scales, for example from 1 O m. to 1 02m. and nanoseconds to years;

. Development of tools to present complex, multi-dimensional data relationships to the human interface. HTRD&E establishes the new paradigm for the researcher who now must interpret data surfaces and not just datapoints.

Design of experiments. Due to the potentially high number of candidates available in HT methodology, the designof the sample library requires rational chemical synthesis or process information to reduce the number of samples andexperiments without increasing the probability for endless searches, false positives, or false negatives.

. New tools will have to be developed to enable the integration of this information into molecular- andproperty-modeling engines. The increased amount of data that can be input into computational engines willrequire a significant increase in speed, bandwidth and storage.

• Advanced, high-speed quantum calculation programs such as Wavefunction, Inc.'s program SPARTAN1° andMSI's program Cerius" will require interfacing with databases and experimental design programs;

• Advances in experimental strategy for dealing with large, diverse chemical spaces, using space-fillingexperimental designs, predictive algorithms, and optimization techniques.

51

FIGuRE 4. THE INTEGRATION OF SOFTWARE SYSTEMS


52

B. HARDWAREThe basic underlying hardware technologies needed for HT library fabrication and analysis are enabling and emergingtechnologies that are fmding applications in other arenas. Broad technical needs have been identified by industry (Table2).

Increasing computer throughput with, for example, massively parallel computers and high bandwidthnetworks;

Micro-machines and micro-reactor technologies (MRT) based on micro-electromechanical systems (MEMS)will address the need for higher library densities to facilitate reduced raw materials costs for library fabricationand economies of scale and modularity in laboratory instrumentation. Construction of solid state materialslibraries currently requires automated onto substrates, using, for example, micro jet, laser ablation, vacuumdeposition (PVD, CVD), or micro-fluidics in a lab-on-chip application. Micro-scale methods have already beencommercialized for health care diagnostics with liquid samples. Preliminary market penetration in the CPI andmaterials arena will be with systems utilizing solution state reaction systems using homogeneous catalysts.

S There is an growth of activity in this area, with links to the HT RD&E conmiunity, especially for drugdiscovery

12 One prototype microreactor is shown in Figure 5.13 There is a growing internationaleffort to understand MRT for distributed manufacturing of chemicals and for high throughput screening.Principal components of MRT are analogous to large-scale machines—reagent mixers, reactors, distribution andseparation. The key technical challenges for MRT are in component integration, development of modularsystems, and development of engineering tools to design and build microscopic systems. Since fluid flow inthese devices is laminar, a significant challenge is in understanding fluid dynamics in these systems. Clearly,the convergence of the microelectronics technologies with micro-reactor technologies will accelerate in the nearfuture, and are being tracked with interest by many parties for their economic potential. The current technologyleaders for chemical process development and high throughput screening are the German Institut firMikrotechnik Mainz'4, and Forschungszentrum Karlsruhe.'5 Research in the United States is centered primarilyat the Pacific Northwest National Laboratory16 and at the Massachusetts Institute of Technology MicrosystemsTechnology Laboratory.'7 A collaboration between DuPont Experimental Station and Massachusetts Institute ofTechnology (MIT) has produced a circuit board-level discovery plant containing multiple socket-borne(interchangeable) catalytic reactors with integrated fluidics control, heat transfer, separation, and mixing for HTdiscovery and process development.'8

FIGuRE 5. PROTOTYPE MIcRo-REAcToR FOR FLUIDIC SYSTEMS FROM MIT

U


FIGuRE 6. MICROREACTOR ASSEMBLY FROM THE INSTITUT FÜR MIKR0TEcHNIK MAINZ

Micro-scale sensors. High throughput methods will drive the development of advanced sensors and sensorarrays. The current technology relies on contact and non-contact methods of characterization and externalcontrol of process conditions. A significant impetus for developing micro-scale sensors has been the U.S.Department of Defense,'9 primarily in response to battlefield detection of chemical and biological warfareagents and portable power generation, and from National Aeronautics and Space Administration (NASA) formicro-robotic space exploration. The CPI sector is moving toward smaller, more integrated sensing devices inprocess control of manufacturing sites. Robotics, next-generation "titer plates", lab-on-a-chip designs, and rapidscanning devices for HT materials innovation will be with application-driven tools targeting specific physicalproperties that can be analyzed at microscopic levels. Automated library processing is especially challenging fornew materials development since samples within a library may require different or non-equilibrium processingparameters across the library. Therefore, with the impetus toward MRT lab-on-chip devices, integration willdrive the development of foundry methods to produce on-chip optical sources (e.g., semi-conductor lasers) aswell as detectors. Finally, the capability for interpretation to bulk characteristics from micro-scale will placeincreasing pressure on computational tools such as QSPR.

TABLE 2: TECHNOLOGY CHALLENGES

SOFTWARE

Technology ChallengeLibrary DesignStatisticsModelingLiterature/Patent Databases

- Development of higher-order designs- (Reviewby NSF is pending)- Query languages; visualization; integration

InformaticsQSPR (structure-property predictions)

Database Query Engines

- Propertyprediction, large-scale correlations, integration intoexperimental design tools-New languages, genetic programs- Inter-operability, enterprise-wide integration

53


HARDWARE

Techno1oy ChallengeScreeningThermal propertiesOptical CharacterizationMechanical PropertiesElectrical PropertiesChemical Properties

2-dimensional or parallel sensors for:- Electrical and Thermal Conductivity- Fluorescence, luminescent properties -Modulus, TensileStrength, Impact resistance- Molecular Weight- Polymer architecture/morphology- Catalyst Turn-over, Selectivity, Conversion

ProcessingControl ofphysical environment - Control oftemperature and pressure over library array with

control over individual sample sites or wells- Sample size--control of interfacial diffusion, mass transportproperties, etc.

DepositionThermally-driven (e-beam, laser, etc.)Laser ablationMicro jetVacuum DepositionRobotic pipetting of nano-aloquots

- Delivery offmite samples or composition spreads ofconsistent, known composition at microscopic sizes:reproducibility.- Reduction of cross-talk of sample properties through diffusion,etc. across substrate surface

Micro-Reactor Technologies- Development of standards for modular componentinterconnections- Process control devices tuned to MRT- Modeling for fluid flow, heat and mass transfer;- Substrates: glass, polymer, metal- Surface micro-machining using wet or dry chemical etching,. . . .laser ablation, mechanical micro-millmg, LIGA processes.

Systems integration

Reactor design

.Manufacturmg processes:

54

V. PROBLEM STATEMENT

The two major technological hurdles are systems integration ofthe hardware and software tools, and the scalability ofmicro-scale results to end-use scale.

A. SYSTEMS INTEGRATION

The transfer oftraditional serial research methodologies to multi-dimensionalparallel methodologies will require theintegration of previously diverse computational and characterization tools for sample library design, sample libraryfabrication, characterization, and informatics.

B. SCALABILITY

The ability to utilize data obtained from libraries of micro-scale samples will require scientific advances in the areas ofinterfacial effects and structure-property relationships.

C. COMMERCIAL CHALLENGES

The major commercial hurdle is validation that HT RD&E methodologies lead to increased productivity, for example interms of increased return on investment (ROI) and higher net present value (NPV) for new products. Time-basedcompetition is driving many industries to increase R&D funding at the expense ofreducing allocations to capital assets.The "new economy" may bring about the acceptance ofnew metrics, for example the rate ofRD&E expenditures vs. therate ofcapital investment in manufacturing assets.2° In addition, the concept ofintellectual property as a valued asset isbecoming more widespread as enterprise-wide knowledge management becomes increasingly viable.

There is a resistance to believe that HT discovery and process development methodologies will impact R&D productivityin most firms. Large capital investments will be made only by the few large companies that can guarantee a retum within


one or two years. The CPI and advanced materials sectors 'will not adopt HT methodology until commercial impacts arevalidated, and the R&D capital assets can be economically utilized. The bulk ofthe smaller firms manufacturingspecialty chemicals and materials would conceivably be a target customer, however their entry will be retarded pendingreduced entry costs. These industries are characterized by R&D budgets that are typically 3% to 4% of sales, with profitmargins in the range of 8% ' As indicated earlier, improved return on RD&E investment and faster time-to-product launch are expected to result in higher overall NPV. Because the CPI and materials sectors generally exhibitcycle times of 7 tolO years for concept-to-product launch, given end-user qualification and manufacturing design andconstruction cycles, HT RD&E projects initiated now will not reach commercial validation until 2003-2005 at theearliest.

Notably, numerous large companies have reported publicly that their HT RD&E efforts have been technically successfulin catalyst discovery and process development, at a typical entry cost of $5M to $20M.2' In order to reach a competitivecritical mass, HT RD&E methodologies must reach industries and companies that cannot make such a substantial capitalinvestment. There is, therefore, a role for the Federal Government to gain a significant impact on the national economyby investing in the development ofnew, lower-cost tools that will ultimately reduce entry costs for industries that fmdimplementation prohibited.

VI. ANALYSIS OF SOLUTION

The implementation ofthis emerging methodology by the advanced materials industry will enable new technologies havingbroad technological and economic impact. The principal drivers are lower development costs, reduced innovation cycletimes, and higher performance materials for industrial and consumer products--to impact the accelerating market needs of"faster-less expensive-better".

Since February 1998, the NIST ATP has been collecting input on the needs ofU.S industry for new techniques andstrategies for HT R&D. The role ofgovernment support in response to industry needs in a relevant time frame, and theeconomic value to taxpayers of a potential investment, are important considerations.

VII. COMMERCIAL OBJECTIVES

A. BACKGROUNDA new paradigm is now taking hold in the RD&E centers ofJapanese, European, and U.S. advanced materials industriesof catalysts, polymers, specialty and fme chemicals, optical materials, and electronic materials.22 High throughputmethods2 for materials RD&E leverages the combinatorial chemistry methodologies developed over the past ten years inthe pharmaceutical industry. The implementation ofthis emerging methodology by materials innovators will enable newtechnologies having broad technological and economic impact as manufacturers strive to meet original equipmentmanufacturer (OEM) demands. The bottom line is that more complex compositions can be created and analyzed in areduced cycle time and at lower cost. The flip side is that this should also be a watershed for basic research in elucidatingnew phenomena as well as in developing the new technology platforms.

Since February 1998, the NIST ATP has been gathering primary and secondary market research through an IndustryWorking Group, by holding public workshops (3 March1998, 28 Nov1998, l6Nov1999), and by a proactive outreacheffort at conferences and workshops. The NIST ATP leveraged this knowledge base to understand U.S. industry needswith respect to the ATP Mission. The ATP's cost-sharing ofhigh-risk industrial research appears to correlate uniquelyto industry's critical success factors.23

There are significant scientific and technological challenges to implement HT methods in advanced materials. Thesechallenges can borrow from advances made in other areas such as drug discovery and bioinformatics, however in mostcases the technology transfer is not straightforward. The transfer of traditional serial research methodologies to parallelmethodologies will require the integration of otherwise diverse computational and characterization tools for librarydesign (statistics, design of experiments), library fabrication (sample deposition, automation), characterization (sensors,automation, data acquisition), and informatics (data handling, visualization, decision tools). The scalability of the resultsobtained will require scientific and technological advances in the areas of interfacial effects and structure-propertyprediction.

55


56

As OEMs and materials manufacturers focus on their core business areas--assembly andmanufacture--the advancedmaterials industries are beginning to outsource discovery andprocess development. This is resulting in the formation of,and dependence on, small firms that are structured as discovery- and/or high throughput methods service providers (the"discovery and process development industry"). There are currently seven of these service firms world-wide, and theircompetencies lie in an ability to focus integrated hardware and software systems toward profitable research areas such ascatalysts and electronic materials. The HT RD&E environment includes materials manufacturers, discovery serviceproviders, and hardware and software tool providers (Figure 7).

B. IMPACT ON INDUSTRY

ATP's support of industrial implementation ofhigh throughput methods could significantly improve the competitivestance ofU.S. industry and provide broad economic impact. Industry representatives have indicated that a reduction inthe time to launch a new catalyst for chemicals manufacture can be reduced approximately two years utilizing HTdiscovery methods. This two-year reduction in break-even point (BEP) may approach $20M for a typical $ lOOMchemical plant. ATP can further decrease the time-to- market through co-funding the high-risk development of new toolsthat would normally be beyond the scope of traditional industrial research, resulting in a higher Net Present Value(NPV).

Cash Flow ($)

FIGURE 8. ATP FUNDING IMPACTS TIME-To-MARKET FOR HIGH-RISK TECHNOLOGIES

FIGURE 7. THE MATERIALS INDUSTRIES HIERARCHY

Break-EvenPoints

— With CombiTraditional

Time


One market factor that might predict the movement of HT RD&E methods is the current level of spending for R&D inthese sectors. Implementation ofHT RD&E technologies is expensive; sufficient R&D funds must be available, andacceptable returns on the investment may be precluded in some industries by their profit structure. Aerospace,automotive, biomedical, telecommunications, and computers are the principal end-users of the chemicals and materialssectors. Semiconductor and electronic hardware sectors are materials and device producers (OEM's), and analyticalinstruments and software sectors provide technology infrastructure. The annual R&D budgets for these finns are anindicator ofthe leverage that HT methods can have: assuming a 5%penetration into the infrastructure ($9.7B in 1997)and supplier sector R&D ($13.4B), approximately $1 .2B in R&D budgets would be assigned to high throughput methodsdevelopment with a potential downstream increase of $905B in sales (pharmaceuticals excluded).

High throughput methods will eventually impact all of the advanced materials industries and high technology industries,therefore the potential impact of the ATP technology cluster "Combinatorial Methods for Materials R&D" could belarge. Funding for two projects awarded in the ATP FY1999 Open Competition is shown below.

$50$45

$40$35

$30$M $25

$20

$15

$10

$5

$0

FIGURE 9. ToTAL ATP AND INDUSTRY INVESTMENT

ATP can also have an important and probably unique role in funding research at NIST in this area through IntramuralFunding. This research addresses fundamental problems in developing new standards and measurement techniques thatwill fmd general use by U.S. industry. Research funded for FY2000 through the ATP Intramural program will focus on:

. developing the infrastructure to support new parallel methodologies and measurement tools that can be tailored toindustrial applications and properties;

. validating new and existing methods and models for small sample sizes analyzed using high throughput approaches;

. validating the applicability ofHT methods to new materials and research problems.

The NIST Measurements and Standards Laboratories (MSL) are uniquely qualified to participate in this area: NIST'sbroad capabilities for scientific research and technology development funding address major challenges of highthroughput methods implementation. ATP is currently funding research in the areas ofpolymer array scaffolds for tissueengineering,24 two-dimensional infrared array detection of inorganic and organic substrates,25 x-ray and microwavemeasurement of dielectric ceramic thin films,26 analysis of dopants in compound semiconductor thin films,27 x-ray

57

Nonlinear DynamicslUOP LLP "Combinatorial Tools and Advanced Data Analysis Methodsfor Heterogeneous Catalysts"$14,715K (ATP) + $15,186 (joint venture, 5 year duration)

GE/Avery-Dennison "Combinatorial Methodology for Coatings Development"$3,127K (ATP) + $3,200K (joint venture, 3 year duration)

2000 2001 2002 2003 2004Fiscal Year


studies ofsupported catalysts,28 and genetic programming ofdata query engines.29 The NIST MSL effort is expected togrow significantly.

C. TECHNOLOGICAL CHALLENGESMaterials manufacturers are utilizing several business scenarios to obtain HT RD&E capabilities: by developing internalcapabilities (e.g., General Electric Corporate R&D); by contracting with service providers having a core competency inhigh throughput discovery methods (e.g., Dow Chemical with Symyx Technologies); or by developing an independentconsortium or alliance partnership with individual tools providers (e.g., Avantium—a Shell spin-off in partnership withthe Dutch government, private investors, and three Dutch universities). These events signal a clear trend that thematerials industry is beginning to dis-integrate their front-end discovery efforts to smaller external entities. There arecurrently seven companies known to be either performing front-end R&D or developing integrated systems for largematerials manufacturers utilizing HT R&D (ref. Table 5 1)

TABLE 5. HIGH THROUGHPUT METHODS PROVIDERS FOR ADVANCED MATERiALS

Parent Company(Miscellaneous)'2

Company(HT Materials R&D)

Financing!Partnerships

Symyx Technologies (Santa Clara) IPO-l 1/20/99Hoechst AGC, Celanesec, BayerAGC, BASFC, B.F. Goodrichc,Dow ChemicalUnilever

ArQule Inc. Alveus Systems (Belmont, MA) N/ABASF hte GmbH (Heidelberg) MPI-Kohlenforschung

BASF, Private fmancingCharybdis Scylla (San Diego) Private financingSRI International CombiCat (Menlo Park) Internal, clientsCatalytica AdvancedTechnologies

Aperion (Mountain View, CA) (Announced June 1999)

Molecular Design (UK) Cambridge Discovery Chemistry(Cambridge, UK)

BP Chemicals (Germany)DSM (Netherlands)Engeihard

Argonaut Technologies Symyx TechnologiesShell Oil Avantium Universities of Delft, Twente,

Eindhoven; the government ofThe Netherlands; GSE, Inc.; othershareholders.

VIII. CONCLUSION

Combinatorial methods have revolutionized drug discovery in the pharmaceutical industry and are poised to do the samein materials science, biotechnology, chemistry, physics and other technological bases. These efforts are needed tomaintain U.S. leadership in chemicals and materials R&D by providing globally competitive increases in efficiency ofmaterials discovery and process optimization. Industry has correlated the impact of increased innovation throughput indifferent areas to broad economic benefits.

HT methods R&D for advanced materials is about four years old.3° Current end users for HT RD&E are the chemicalprocess industry (catalyst discovery and chemical process development), specialty polymers, and electronic materials(e.g., inorganic dielectric and phosphor materials discovery) intr31 One would expect that, as the barrier to entryis reduced by lower cost tools and validation ofthe methodologies, lower-valued applications,32 for example, commoditymaterials, will become targeted. Key commercial and technology drivers with regard to implementation of HT RD&Emethods are shown in Table 6.

58


TABLE 6. MAPPING TEcHNIcALCHALLENGES TO ECONOMIC BENEFITS

Application Areas Technical Challenges Impacted Products Economic BenefitsIndustrial Chemicalsand Monomers

• Faster catalyst screening.• Capability of screeningextremely diversecombinations of catalystingredients.

• Industrialchemicals• EngineeringThermoplastics• Other plastics

• Lower cost• Lower energy usage• New products based onnewly affordable rawmaterials

Polymers• Catalysts• Polymer blends• Surface modifiers

• Faster screening• Process optimization

• Engineeringplastics• Thermoplastics

• New products• New markets such asautomotive glazing• Reduced domesticenergy consumption.

Ceramics • Thermal barrier coatingoptimization• Electronic properties• Higher strength

• Aircraft engines,advanced powermachines,• Conductors, semi-conductors, dielectrics• Machine tools

• Higher enginetemperatures, increasedservice life, and reduction ofdowntime

• Higher componentdensities and speeds• Machining of newalloys, increased productivity

Advances in both hardware and software will be required to implement HT RD&E methods more broadly. The basicunderlying software technology must be capable defming profitable experimental spaces and visualizing complex datarelationships, and of correlating target materials with prope.rties to permit data-base queries from a broad spectrum ofdata mining engines and the development of structure-property relationships. This requires interfacing with datavisualization tools at the back end and statistical experimental design engines on the front end while remaining compliantwith enterprise-wide systems for knowledge management and maintaining control of experimental hardware.

Through competitive proposal submissions from industry teams to the ATP Open Competitions, the ATP can reduce thecost of implementing high throughput RD&E by cost-sharing the high technical-risk aspects of developing new, generictools so that cost-sensitive industrial sectors will be able to invest in this new capability. Successful proposals to the ATPhave the potential to facilitate the emergence ofnew technology platforms by stimulating the development ofpartnerships; it can spur the fusion ofnew technologies for subsequent diffusion to other areas; it can facilitate theintegration of complex systems; it can help the integration of otherwise diverse technologies.

59


60

REFERENCES

1 Certain commercial equipment, instruments, or materials are identified in this paper in order to specify theexperimental procedure adequately. Such identification is not intended to imply recommendation or endorsement by theNational Institute of Standards and Technology, nor is it intended to imply that the materials or equipment identified arenecessarily the best available for the purpose.2 The Global Competitiveness Report, 1996, World Economic Forum, Geneva, Switzerland WORLD BANK, From Planto Market: WorldDevelopment Report 1996, NSF, Science and Engineering Indicators, 1996.3 Jandeleit B, Schaefer DJ, Powers IS, Turner HW, Weinberg WH "Combinatorial Materials Science and Catalysis,"Angewandte Chemie-International Edition 1999, 38: (17) 2495-2532.4 Five industry White Papers have been received in this area by the ATP. In addition, pre- and full proposals weresubmitted to the FY1999 ATP Open Competition with two substantial awards made; significant requests have been madeto the AlP Combinatorial Methods web site; ATP public Workshops held in 1998 and 1999 drew approximately 75 and50 representatives, respectively; and continued interest is being expressed in workshops attended by ATP staff.5

Cawse, J., "Information Based Strategies for Combinatorial and High Throughput Materials Development: A WhitePaper," General Electric Co. Corporate R&D, 1 1/1999.6 Technologies, Inc., Santa Clara, CA. "Symyx's patent estate today includes eight issued U.S. patents, as well asmore than 155 patent applications in the U.S. and abroad. 5ymyx's comprehensive intellectual property portfolioincludes patents, patent applications and trade secrets directed toward composition of matter, libraries, synthesis and highthroughput screening methodologies, instrumentation and software" ref. http://www.symyx.com; Patents issued: U.S.Patent 6,030,917, U.S. Patent 6,004,617 and U.S. Patent 5,985,356.7Koinuma H., Koida T, Ohnishi T, et al. "Parallel fabrication ofartificially designed superlattices by combinatorial laserMBE," Appl. Phys. A-Mater. 1999, 69: 529-53 1; Matsumoto Y, Murakami M, Jm ZW, et al., "Combinatorial lasermolecular beam epitaxy (MBE) growth of Mg-Zn-O alloy for band gap engineerg," Japan. I Appl. Phys. 1999, 2 38:(6AB) L603-L605.8For-profit organizations (Knowledge Foundation, CHI, The Catalyst Group, etc.) as well as non-profit organizations

(American Chemical Society, American Physical SocieW, Materials Research Society, Association of IndustrialChemical Engineers, etc.) have held extensive public discussions ofthis area. Consultation ofthe corresponding websites will re-direct interested parties to specific events. In addition, the AlP has held three public workshops:http://www.atp.nist.gov/www/ccmr/ccmroff.htm.9

Symyx Technologies press release, April 7, 1999 and Prospectus filed with the U.S. Securities and ExchangeCommission on November 17, 1999 prior to their Initial Public Offering (http://www.sec.gov).10

http://www.waveftinction.com11

http://www.msi.com12 Ref. K.F. Jensen et al. Massachusetts Institute of Technology; R.S. Wegeng et al. Pacific Northwest NationalLaboratories; W. Ehrhard et a!. Instut fir Mikroteknik Mainz; J.F. Ryley et al. DuPont, etc.13

Catalytic micro-reactor with integrated heaters, temperature and flow sensors. Reactants enter at the ends ofthe bar ofthe T-shaped reactor, mix and react in the heated channel over a catalyst, and exit at the end ofthe channel. Ref Gleasonet al. , http://web.mit.edu/cheme/www/People/Faculty.'4W. Ehrfeld, 1MM, Mainz, Germany.15

http://www.fzk.de/pmt/16 R.S. Wegeng, PNINL, Redland, WA.17 K.F. Jensen, MIT, Cambridge, MA, http://www-mtl.mit.edulmtthome/.18 DL. Quiram et al. Proceedings ofthe 2000 AIChE Spring Meeting, 4" International Conference on MicroreactionTechnology, 06 March 2000.19 DARPA MicroFLUMES Program, http://www.darpa.mil/MTO/mFlumes/index.html.20 D.A. Hicks, "Time Wars: Is There a Financial Undertow from Accelerating Technical Advance?" R-TManagement2000, p. 34-39.21 Forexample, trade journals and presentations at recent public meetings include statements by UOP LLC, GeneralElectric, DuPont, Dow Chemical, BASF, Shell, etc.22

Fairley P., Chemical Week, August 11, 1999, pp. 27-33, and references therein.23 Ref SWOT Analysis of U.S. industry vis a vis combinatorial methods, Appendix Figure 6http://www.atp.nist.gov/www/ccmr/public.pdf24Amis E. et al. Polymers Division, Materials Science and Engineering Laboratory, NIST.25Heilweil E. et al. Optical Division, Physics Laboratory, NIST.


26 Stranick S. et al. Microstructure Characterization Division, Chemical Sciences and Technology Laboratory, NIST; D.Kaiser et al. Ceramics Division, Materials Science and Engineering Laboratory, NIST.27 Handwerker C. et al. Metallurgy Division, Materials Science and Engineering Laboratory, NIST.28 Fischer D., Ceramics Division, Materials Science and Engineering Laboratory, NIST.29 • Devaney and J. Hagadorn, High Performance Computing Division, Information Technology Laboratory, NIST.°

XiangXD, Sun XD, Briceno G, Lou YL, Wang KA, Chang HY, Wallacefreedman WG, Chen SW, Schultz PG, ACombinatorial Approach To Materials Discovery," Science 1995,, 268: (5218) 1738-1740.31 Current age depends on what is considered combinatorial chemistry. M. Geisen, the "father of combinatorialchemistry" published his first research paper in the 1980's, and Merrifield's Resin has long been used for separations,but the methods did not become accepted and automated until the early 1990's. Ref I Combinatorial Chemistry 1999,1(1), 3-10.32 Combinatorial methods are most suitable to compositionally-driven materials previously discovered (laboriously)using purely empirical methods and serial methodologies.

61


high throughput methodologies for chemicals and materials rd&e · the u.s. chemical process...

Documents